12 January 2015
Department of Statistics
University of Warwick
Coventry
CV4 7AL
B Galvin
Group Chief Executive Officer
USS
Royal Liver Building
Liverpool
L3 1PY
Your ref: BG/JMR
Dear Mr Galvin
Thank you for your letter of 18 December 2014. We have numbered the paragraphs of
your letter, and use this to refer to it in our response below.
Most importantly, a conspicuous omission from your letter is an attempt to address the
issue of the wild fluctuation in valuations (our letter, page 2, second last paragraph).
We would appreciate a response as to why the USS Trustee does not recognise such
instability as an indication that the methods of estimation are clearly not fit for purpose
and therefore challenge the methods.
1. We provided comments to the University of Warwick as part of the response to the
Universities UK response.
2. This fails to address the point we made. We did not refer to the USS’s previous or
historic approach to valuation assumptions, but rather to the USS fund’s historic
performance. We are puzzled as to why this historic performance is ignored.
3. We, along with many others, disagree with these claims. QE is producing a demanddriven yield decrease across the range of gilts maturities because the Bank of
England's operations take place across the full range of gilts and maturities. As we are
sure you know, as large proportions of the gilts issued by the government (DMO) are
being bought by the Bank of England in QE operations, even the long-dated gilts do
not reflect market expectations for yields, but rather short-term considerations about
supply and demand. With regard to `second guessing' the economic markets, as we
are sure you are aware, the “taper tantrum” in 2013 demonstrates how gilt yields rise
dramatically in response to the possibility of QE ending, while the Bank of England has
indicated an intention to cease QE on a timescale which is very short in comparison to
long gilt maturities.
4. We are most interested by this paragraph and are perplexed as to why this was not
presented clearly (or at all) in public communications. We would welcome receiving the
details, particularly the interpretation of `realism’ and the `appropriate margin for
prudence' in relation to actual performance of the various asset classes.
5, 6 & 7. (Replies to point 2 in our November letter.) We note that you describe the RPI
inflation assumption as an objective measure.
We note that you have not acknowledged that public estimates of the inflation risk
premium are in the range 0.5%-1% for 5 to 10 year bond maturities (page 3 of
referenced paper), and must logically be higher for longer maturities (since the
cumulative uncertainty about yields has to be greater). This is supported by historic
data from the Bank of England.
With respect to the last sentence of para 6: the actions of USS can have no effect on
the market-implied Inflation Risk Premium; we do accept that you intend to decrease
the yield from USS assets.
The latest estimate of CPI from the Bank of England Monetary committee for the year
2017 is 1.91% and the present target is 2%, which implies that RPI should lie between
2.7% and 2.9%, in contrast to the 3.4% assumed by USS (3.6% - 0.2%).
8. With regard to salary growth, we note that you choose to ignore the most recent
three years of salary data: is this ignored because it is inconvenient? Even with this
exclusion, you cite general pay growth as RPI +0.7%, not +1.0%.
We would welcome an explanation of why the trustee is minded to assume 3.6%+1.0%
= 4.6% for future salary instead of the evidence-based 2.9%+0% = 2.9%.
9. We are puzzled, indeed concerned, by your inclusion of the first sentence of
paragraph 9 as, although it is factually correct, it is entirely irrelevant to the conclusions
of any given actuarial valuation. It is the conclusions of the April 2014 valuation and the
corresponding estimates of liability which drive the present discussion.
If in 1990, the assumption of RPI +1% had been used in the 1990 valuation, then the
estimate of liability for 2014 would have been overstated by 25%: this is why the
expectation of future increases is important. In this case we do know that the liabilities
would have been `overstated', i.e. that the estimate of liabilities is seriously biased. The
equivalent inference is that an assumption of RPI + 1% in April 2014 results in an
overstatement of liabilities in 2034.
10. We referred not to one year of observed life expectancy, but to the most recent
working papers published by CMI: wp63 and wp69, which refer to recent years.
CMIwp69 also stresses the sensitivity to the assumptions on the rate of mortality
improvement. Did you wish to imply that the most recent CMI reports are not valid?
With regard to supporting data, we refer you to CMI wp39. Figure 6.1, for example,
clearly shows the so-called 'golden cohort', born 1927-1936, who have annual mortality
improvements which are higher than those born later. The observed data for those
born in the 1960s includes negative annual mortality improvements.
11. Your comment on circularity is tangential, at best, to the points which we made. If
the assumptions are determined (driven) by the target to be achieved, the assumptions
include that target, as the target is an estimate derived from assumptions.
We refer to the desire to reduce the return on the investments in order to decrease the
variability in estimates of liability, i.e. `risk' is expressed as `variance', not as potential
loss to members of USS. The assumption that future returns are dominated by gilts
both reduces variability and increases the estimated deficit.
12. You have conflated the period over which estimated excess liabilities need to be
funded with the period over which the funders will tolerate any variability in contribution
rate. The 40 years might refer to the technical average duration of fund. We
understand that the Ernst & Young covenant assessment confirmed the enduring
nature and solvency of the higher education sector to and beyond 20 years.
In the light of your statement about benefit design, we would like to have your
comments on these remarks from UUK’s 22 October submission to the JNC: `It will be
noted that in looking at their Tests, the Trustees take no account of any Defined
Contribution benefits that are introduced. These pose no obligation risk on the
employers – in effect, risk is borne fully by members. As such, the Trustees exhibit a
marked preference for DC benefits over DB benefits.'
The USS publicity officer who spoke at Warwick University (9-11 December 2014)
also stated that the aim of the move to DC is to transfer all risk from the employers to
the members of USS. Since you have agreed that the sole obligation of the Trustees is
to the scheme's beneficiaries, it seems highly invidious that they were enthusiastically
agitating for such a change.
Yours sincerely,
Prof. J L Hutton
Prof. S D Jacka
Cc:
Professor Steven Haberman, S.Haberman@city.ac.uk
Professor Simon Wood, s.wood@bath.ac.uk.
Professor Sir Christopher Snowden, President, UUK, h.staveley@surrey.ac.uk.
The Pensions Regulator, customersupport@tpr.gov.uk
Dr. Ros Altmann CBE, ros@rosaltmann.com